The sensitivity of a numerical prognosis to moisture detail in the initial state

The sensitivity of the ARPE model prognosis to changes in the initial specification of moisture distribution is demonstrated for a heavy rainfall situation over south-western Western Australia. it is shown that by manually modifying the objective moisture analyses, using both surface data and satellite imagery, improvements in both mean sea level pressure and precipitation forecasts are achieved. It is shown that the improvement in the precipitation forecast over southwest Western Australia was due to the changed distribution of moisture associataI with a middle-level cloud layer, while the improvement in the forecast movement of a low pressure system was caused by the interaction of the model's cumulus parameterisation scheme with the changed moisture distribution of a deep cloud mass associated with the low.